Nagios Support Forum

We monitor a few hundred WAN connections at remote sites and generate a good amount of alerts to our pseudo NOC. Everytime someone clicks on the "Notifications" menu in the main window NagiosXI comes to a crawl and mysql utilization spikes. We don't really use this menu, is there a way to remove it?

You can certainly look in the php for this, however my guess is that this is a protected file that cannot be modified as it is a part of the main xi interface. Any changes that are made, would also need to be done after each and every upgrade too. It would probably be a better idea to try and diagnose why opening that menu is so slow.

Well this gives me a good reason to organize our contacts to point to a distribution group instead of individual accounts for the majority of our agents. Any advice on where to start diagnosing the slow down? I'm thinking because we generate so many alerts the slowdown is related to pulling so many events. I'm not really a mysql guy, any advice is appreciated.

How many notifications are being shown when it is opened? It may be that your users are trying to view a giant list of notifications which is forcing Nagios to open up loads of archived nagios logs and reap that date, then display it. Have you tried limiting it to "past 24 hours" or are you already just loading that?

I'm not currently limiting the amount displayed, I believe that you're correct in that we're displaying too many notifications. Without clicking into that menu, is there a way to limit how many are displayed?

Possibly by editing the entry in postgres. I will see what I can find. You an also truncate the table, or decrease the max age of the table: Admin --> Performance Settings --> Databases tab --> NDOUtils --> Max Log Entries Age. Change it to 30 and then update settings. Let us know if you are still having issues with the notifications menu option. If so, I will look at ways to alter the display quantity from postgres.

Just to clarify should I limit the log age or the notification age for this particular issue? Logs would be user auth logs etc, correct?

Thanks for the advice guys I really appreciate the level of support I've gotten when I've needed it.

Ehamby wrote:Just to clarify should I limit the log age or the notification age for this particular issue? Logs would be user auth logs etc, correct?

Good catch, change the notifications age to 30 days.

Ehamby wrote: Thanks for the advice guys I really appreciate the level of support I've gotten when I've needed it.

You are most welcome

Well, it did help cut the load time significantly but in the end it still causes Nagios to become unresponsive for anyone using it during the load (was 10-15 minutes, now its ~5). I'm not sure if the db would clear those entries right after I changed and applied that setting; I loaded the notification menu about 30 seconds after applying the change. Below is an image showing the CPU load on the VM during the event. I suppose its expected, but I figured I'd attach it just to give more insight.
One thing I'm going to do is remove the contacts for these users from the contact group, and specify a single Exchange distribution group to distribute the notifications to the users. That way instead of 20 notifications for each event it'll cut it down to ~5 (IT as individual contacts and NOC staff as the distro). Probably should have done that from the get go but cest le vie.

It would make sense that there is some load increase between php and mysql as those notifications need to be indexed, but I agree that does seem to be quite a lot. That said, I wouldn't expect the logs to get cleaned up for at least an hour or so, depending on when the cleanup job is run. Also I think your distribution list idea is perfect, and should significantly lower this usage even more, especially over 1-2 weeks as those logs get quite a bit slimmer.